skill-name

name	<skill-name>
description	<short description>
disable-model-invocation	true
allowed-tools	Read, Grep, Glob
argument-hint	[args]

Purpose

Describe what this skill does and when to use it.

Inputs

• List required inputs

• Template paths are resolved from ${AGENTS_SKILLS_ROOT}/<skill-name>/.

Steps

1. Provide clear, ordered instructions.
2. Keep actions explicit and verifiable.
3. If this skill uses external command specs, reference them with placeholder paths: ${AGENTS_SKILLS_ROOT}/<skill-name>/commands/<file-name>.md.
   • Avoid fragile repo-root relative paths like ./commands/<file-name>.md.
4. If this skill uses templates/... relative paths, resolve them from ${AGENTS_SKILLS_ROOT}/<skill-name>/.
5. If this skill needs executable scripts, place them under ${AGENTS_SKILLS_ROOT}/<skill-name>/scripts/ (source path in repo: .agents/skills/<skill-name>/scripts/).
6. Split intelligence vs determinism:
   • Use the LLM for intent parsing and decision-making.
   • Use scripts in tools/ for parsing, transforms, formatting, and API calls.
   • Define strict input/output contracts for each script.
7. If the skill must be discoverable by OpenCode, run pnpm skills:register.

Deterministic Logic (Scripts)

• List scripts in tools/ and their exact input/output schema.
• Prefer JSON in/out for machine-readability.
• Keep scripts idempotent and testable.

Output

• Expected output format or success criteria.